use of org.apache.hadoop.hive.serde2.io.DoubleWritable in project hive by apache.
the class VectorColumnAssignFactory method buildObjectAssign.
public static VectorColumnAssign buildObjectAssign(VectorizedRowBatch outputBatch, int outColIndex, PrimitiveCategory category) throws HiveException {
VectorColumnAssign outVCA = null;
ColumnVector destCol = outputBatch.cols[outColIndex];
if (destCol == null) {
switch(category) {
case VOID:
outVCA = new VectorLongColumnAssign() {
// This is a dummy assigner
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
// This is no-op, there is no column to assign to and val is expected to be null
assert (val == null);
}
};
break;
default:
throw new HiveException("Incompatible (null) vector column and primitive category " + category);
}
} else if (destCol instanceof LongColumnVector) {
switch(category) {
case BOOLEAN:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
BooleanWritable bw = (BooleanWritable) val;
assignLong(bw.get() ? 1 : 0, destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case BYTE:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
ByteWritable bw = (ByteWritable) val;
assignLong(bw.get(), destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case SHORT:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
ShortWritable bw = (ShortWritable) val;
assignLong(bw.get(), destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case INT:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
IntWritable bw = (IntWritable) val;
assignLong(bw.get(), destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case LONG:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
LongWritable bw = (LongWritable) val;
assignLong(bw.get(), destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case TIMESTAMP:
outVCA = new VectorTimestampColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
assignTimestamp((TimestampWritable) val, destIndex);
}
}
}.init(outputBatch, (TimestampColumnVector) destCol);
break;
case DATE:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
DateWritable bw = (DateWritable) val;
assignLong(bw.getDays(), destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case INTERVAL_YEAR_MONTH:
outVCA = new VectorLongColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
HiveIntervalYearMonthWritable bw = (HiveIntervalYearMonthWritable) val;
assignLong(bw.getHiveIntervalYearMonth().getTotalMonths(), destIndex);
}
}
}.init(outputBatch, (LongColumnVector) destCol);
break;
case INTERVAL_DAY_TIME:
outVCA = new VectorIntervalDayTimeColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
HiveIntervalDayTimeWritable bw = (HiveIntervalDayTimeWritable) val;
assignIntervalDayTime(bw.getHiveIntervalDayTime(), destIndex);
}
}
}.init(outputBatch, (IntervalDayTimeColumnVector) destCol);
break;
default:
throw new HiveException("Incompatible Long vector column and primitive category " + category);
}
} else if (destCol instanceof DoubleColumnVector) {
switch(category) {
case DOUBLE:
outVCA = new VectorDoubleColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
DoubleWritable bw = (DoubleWritable) val;
assignDouble(bw.get(), destIndex);
}
}
}.init(outputBatch, (DoubleColumnVector) destCol);
break;
case FLOAT:
outVCA = new VectorDoubleColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
FloatWritable bw = (FloatWritable) val;
assignDouble(bw.get(), destIndex);
}
}
}.init(outputBatch, (DoubleColumnVector) destCol);
break;
default:
throw new HiveException("Incompatible Double vector column and primitive category " + category);
}
} else if (destCol instanceof BytesColumnVector) {
switch(category) {
case BINARY:
outVCA = new VectorBytesColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
BytesWritable bw = (BytesWritable) val;
byte[] bytes = bw.getBytes();
assignBytes(bytes, 0, bw.getLength(), destIndex);
}
}
}.init(outputBatch, (BytesColumnVector) destCol);
break;
case STRING:
outVCA = new VectorBytesColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
Text bw = (Text) val;
byte[] bytes = bw.getBytes();
assignBytes(bytes, 0, bw.getLength(), destIndex);
}
}
}.init(outputBatch, (BytesColumnVector) destCol);
break;
case VARCHAR:
outVCA = new VectorBytesColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
// We store VARCHAR type stripped of pads.
HiveVarchar hiveVarchar;
if (val instanceof HiveVarchar) {
hiveVarchar = (HiveVarchar) val;
} else {
hiveVarchar = ((HiveVarcharWritable) val).getHiveVarchar();
}
byte[] bytes = hiveVarchar.getValue().getBytes();
assignBytes(bytes, 0, bytes.length, destIndex);
}
}
}.init(outputBatch, (BytesColumnVector) destCol);
break;
case CHAR:
outVCA = new VectorBytesColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
// We store CHAR type stripped of pads.
HiveChar hiveChar;
if (val instanceof HiveChar) {
hiveChar = (HiveChar) val;
} else {
hiveChar = ((HiveCharWritable) val).getHiveChar();
}
byte[] bytes = hiveChar.getStrippedValue().getBytes();
assignBytes(bytes, 0, bytes.length, destIndex);
}
}
}.init(outputBatch, (BytesColumnVector) destCol);
break;
default:
throw new HiveException("Incompatible Bytes vector column and primitive category " + category);
}
} else if (destCol instanceof DecimalColumnVector) {
switch(category) {
case DECIMAL:
outVCA = new VectorDecimalColumnAssign() {
@Override
public void assignObjectValue(Object val, int destIndex) throws HiveException {
if (val == null) {
assignNull(destIndex);
} else {
if (val instanceof HiveDecimal) {
assignDecimal((HiveDecimal) val, destIndex);
} else {
assignDecimal((HiveDecimalWritable) val, destIndex);
}
}
}
}.init(outputBatch, (DecimalColumnVector) destCol);
break;
default:
throw new HiveException("Incompatible Decimal vector column and primitive category " + category);
}
} else {
throw new HiveException("Unknown vector column type " + destCol.getClass().getName());
}
return outVCA;
}
use of org.apache.hadoop.hive.serde2.io.DoubleWritable in project hive by apache.
the class VectorDeserializeRow method convertRowColumn.
/**
* Convert one row column value that is the current value in deserializeRead.
*
* We deserialize into a writable and then pass that writable to an instance of VectorAssignRow
* to convert the writable to the target data type and assign it into the VectorizedRowBatch.
*
* @param batch
* @param batchIndex
* @param logicalColumnIndex
* @throws IOException
*/
private void convertRowColumn(VectorizedRowBatch batch, int batchIndex, int logicalColumnIndex) throws IOException {
final int projectionColumnNum = projectionColumnNums[logicalColumnIndex];
Writable convertSourceWritable = convertSourceWritables[logicalColumnIndex];
switch(sourceCategories[logicalColumnIndex]) {
case PRIMITIVE:
{
switch(sourcePrimitiveCategories[logicalColumnIndex]) {
case VOID:
convertSourceWritable = null;
break;
case BOOLEAN:
((BooleanWritable) convertSourceWritable).set(deserializeRead.currentBoolean);
break;
case BYTE:
((ByteWritable) convertSourceWritable).set(deserializeRead.currentByte);
break;
case SHORT:
((ShortWritable) convertSourceWritable).set(deserializeRead.currentShort);
break;
case INT:
((IntWritable) convertSourceWritable).set(deserializeRead.currentInt);
break;
case LONG:
((LongWritable) convertSourceWritable).set(deserializeRead.currentLong);
break;
case TIMESTAMP:
((TimestampWritable) convertSourceWritable).set(deserializeRead.currentTimestampWritable);
break;
case DATE:
((DateWritable) convertSourceWritable).set(deserializeRead.currentDateWritable);
break;
case FLOAT:
((FloatWritable) convertSourceWritable).set(deserializeRead.currentFloat);
break;
case DOUBLE:
((DoubleWritable) convertSourceWritable).set(deserializeRead.currentDouble);
break;
case BINARY:
if (deserializeRead.currentBytes == null) {
LOG.info("null binary entry: batchIndex " + batchIndex + " projection column num " + projectionColumnNum);
}
((BytesWritable) convertSourceWritable).set(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesLength);
break;
case STRING:
if (deserializeRead.currentBytes == null) {
throw new RuntimeException("null string entry: batchIndex " + batchIndex + " projection column num " + projectionColumnNum);
}
// Use org.apache.hadoop.io.Text as our helper to go from byte[] to String.
((Text) convertSourceWritable).set(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesLength);
break;
case VARCHAR:
{
// that does not use Java String objects.
if (deserializeRead.currentBytes == null) {
throw new RuntimeException("null varchar entry: batchIndex " + batchIndex + " projection column num " + projectionColumnNum);
}
int adjustedLength = StringExpr.truncate(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesLength, maxLengths[logicalColumnIndex]);
((HiveVarcharWritable) convertSourceWritable).set(new String(deserializeRead.currentBytes, deserializeRead.currentBytesStart, adjustedLength, Charsets.UTF_8), -1);
}
break;
case CHAR:
{
// that does not use Java String objects.
if (deserializeRead.currentBytes == null) {
throw new RuntimeException("null char entry: batchIndex " + batchIndex + " projection column num " + projectionColumnNum);
}
int adjustedLength = StringExpr.rightTrimAndTruncate(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesLength, maxLengths[logicalColumnIndex]);
((HiveCharWritable) convertSourceWritable).set(new String(deserializeRead.currentBytes, deserializeRead.currentBytesStart, adjustedLength, Charsets.UTF_8), -1);
}
break;
case DECIMAL:
((HiveDecimalWritable) convertSourceWritable).set(deserializeRead.currentHiveDecimalWritable);
break;
case INTERVAL_YEAR_MONTH:
((HiveIntervalYearMonthWritable) convertSourceWritable).set(deserializeRead.currentHiveIntervalYearMonthWritable);
break;
case INTERVAL_DAY_TIME:
((HiveIntervalDayTimeWritable) convertSourceWritable).set(deserializeRead.currentHiveIntervalDayTimeWritable);
break;
default:
throw new RuntimeException("Primitive category " + sourcePrimitiveCategories[logicalColumnIndex] + " not supported");
}
}
break;
default:
throw new RuntimeException("Category " + sourceCategories[logicalColumnIndex] + " not supported");
}
/*
* Convert our source object we just read into the target object and store that in the
* VectorizedRowBatch.
*/
convertVectorAssignRow.assignConvertRowColumn(batch, batchIndex, logicalColumnIndex, convertSourceWritable);
}
use of org.apache.hadoop.hive.serde2.io.DoubleWritable in project presto by prestodb.
the class OrcFileRewriter method uncompressedSize.
private static int uncompressedSize(Object object) throws IOException {
if (object instanceof OrcStruct) {
OrcStruct struct = (OrcStruct) object;
int size = 0;
for (int i = 0; i < struct.getNumFields(); i++) {
size += uncompressedSize(getFieldValue(struct, i));
}
return size;
}
if ((object == null) || (object instanceof BooleanWritable)) {
return SIZE_OF_BYTE;
}
if (object instanceof LongWritable) {
return SIZE_OF_LONG;
}
if (object instanceof DoubleWritable) {
return SIZE_OF_DOUBLE;
}
if (object instanceof Text) {
return ((Text) object).getLength();
}
if (object instanceof BytesWritable) {
return ((BytesWritable) object).getLength();
}
if (object instanceof List<?>) {
int size = 0;
for (Object element : (Iterable<?>) object) {
size += uncompressedSize(element);
}
return size;
}
if (object instanceof Map<?, ?>) {
int size = 0;
for (Map.Entry<?, ?> entry : ((Map<?, ?>) object).entrySet()) {
size += uncompressedSize(entry.getKey());
size += uncompressedSize(entry.getValue());
}
return size;
}
throw new IOException("Unhandled ORC object: " + object.getClass().getName());
}
use of org.apache.hadoop.hive.serde2.io.DoubleWritable in project hive by apache.
the class DruidSerDe method deserialize.
@Override
public Object deserialize(Writable writable) throws SerDeException {
final DruidWritable input = (DruidWritable) writable;
final List<Object> output = Lists.newArrayListWithExpectedSize(columns.length);
for (int i = 0; i < columns.length; i++) {
final Object value = input.getValue().get(columns[i]);
if (value == null) {
output.add(null);
continue;
}
switch(types[i].getPrimitiveCategory()) {
case TIMESTAMP:
output.add(new TimestampWritable(Timestamp.valueOf(ZonedDateTime.ofInstant(Instant.ofEpochMilli(((Number) value).longValue()), tsTZTypeInfo.timeZone()).format(DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss")).toString())));
break;
case TIMESTAMPLOCALTZ:
output.add(new TimestampLocalTZWritable(new TimestampTZ(ZonedDateTime.ofInstant(Instant.ofEpochMilli(((Number) value).longValue()), ((TimestampLocalTZTypeInfo) types[i]).timeZone()))));
break;
case BYTE:
output.add(new ByteWritable(((Number) value).byteValue()));
break;
case SHORT:
output.add(new ShortWritable(((Number) value).shortValue()));
break;
case INT:
output.add(new IntWritable(((Number) value).intValue()));
break;
case LONG:
output.add(new LongWritable(((Number) value).longValue()));
break;
case FLOAT:
output.add(new FloatWritable(((Number) value).floatValue()));
break;
case DOUBLE:
output.add(new DoubleWritable(((Number) value).doubleValue()));
break;
case DECIMAL:
output.add(new HiveDecimalWritable(HiveDecimal.create(((Number) value).doubleValue())));
break;
case CHAR:
output.add(new HiveCharWritable(new HiveChar(value.toString(), ((CharTypeInfo) types[i]).getLength())));
break;
case VARCHAR:
output.add(new HiveVarcharWritable(new HiveVarchar(value.toString(), ((VarcharTypeInfo) types[i]).getLength())));
break;
case STRING:
output.add(new Text(value.toString()));
break;
case BOOLEAN:
output.add(new BooleanWritable(Boolean.valueOf(value.toString())));
break;
default:
throw new SerDeException("Unknown type: " + types[i].getPrimitiveCategory());
}
}
return output;
}
use of org.apache.hadoop.hive.serde2.io.DoubleWritable in project nifi by apache.
the class TestConvertAvroToORC method test_onTrigger_complex_record.
@Test
public void test_onTrigger_complex_record() throws Exception {
Map<String, Double> mapData1 = new TreeMap<String, Double>() {
{
put("key1", 1.0);
put("key2", 2.0);
}
};
GenericData.Record record = TestNiFiOrcUtils.buildComplexAvroRecord(10, mapData1, "DEF", 3.0f, Arrays.asList(10, 20));
DatumWriter<GenericData.Record> writer = new GenericDatumWriter<>(record.getSchema());
DataFileWriter<GenericData.Record> fileWriter = new DataFileWriter<>(writer);
ByteArrayOutputStream out = new ByteArrayOutputStream();
fileWriter.create(record.getSchema(), out);
fileWriter.append(record);
// Put another record in
Map<String, Double> mapData2 = new TreeMap<String, Double>() {
{
put("key1", 3.0);
put("key2", 4.0);
}
};
record = TestNiFiOrcUtils.buildComplexAvroRecord(null, mapData2, "XYZ", 4L, Arrays.asList(100, 200));
fileWriter.append(record);
fileWriter.flush();
fileWriter.close();
out.close();
Map<String, String> attributes = new HashMap<String, String>() {
{
put(CoreAttributes.FILENAME.key(), "test");
}
};
runner.enqueue(out.toByteArray(), attributes);
runner.run();
runner.assertAllFlowFilesTransferred(ConvertAvroToORC.REL_SUCCESS, 1);
// Write the flow file out to disk, since the ORC Reader needs a path
MockFlowFile resultFlowFile = runner.getFlowFilesForRelationship(ConvertAvroToORC.REL_SUCCESS).get(0);
assertEquals("CREATE EXTERNAL TABLE IF NOT EXISTS complex_record " + "(myInt INT, myMap MAP<STRING, DOUBLE>, myEnum STRING, myLongOrFloat UNIONTYPE<BIGINT, FLOAT>, myIntList ARRAY<INT>)" + " STORED AS ORC", resultFlowFile.getAttribute(ConvertAvroToORC.HIVE_DDL_ATTRIBUTE));
assertEquals("2", resultFlowFile.getAttribute(ConvertAvroToORC.RECORD_COUNT_ATTRIBUTE));
assertEquals("test.orc", resultFlowFile.getAttribute(CoreAttributes.FILENAME.key()));
byte[] resultContents = runner.getContentAsByteArray(resultFlowFile);
FileOutputStream fos = new FileOutputStream("target/test1.orc");
fos.write(resultContents);
fos.flush();
fos.close();
Configuration conf = new Configuration();
FileSystem fs = FileSystem.getLocal(conf);
Reader reader = OrcFile.createReader(new Path("target/test1.orc"), OrcFile.readerOptions(conf).filesystem(fs));
RecordReader rows = reader.rows();
Object o = rows.next(null);
assertNotNull(o);
assertTrue(o instanceof OrcStruct);
TypeInfo resultSchema = TestNiFiOrcUtils.buildComplexOrcSchema();
StructObjectInspector inspector = (StructObjectInspector) OrcStruct.createObjectInspector(resultSchema);
// Check some fields in the first row
Object intFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("myInt"));
assertTrue(intFieldObject instanceof IntWritable);
assertEquals(10, ((IntWritable) intFieldObject).get());
Object mapFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("myMap"));
assertTrue(mapFieldObject instanceof Map);
Map map = (Map) mapFieldObject;
Object mapValue = map.get(new Text("key1"));
assertNotNull(mapValue);
assertTrue(mapValue instanceof DoubleWritable);
assertEquals(1.0, ((DoubleWritable) mapValue).get(), Double.MIN_VALUE);
mapValue = map.get(new Text("key2"));
assertNotNull(mapValue);
assertTrue(mapValue instanceof DoubleWritable);
assertEquals(2.0, ((DoubleWritable) mapValue).get(), Double.MIN_VALUE);
}
Aggregations